Ryan Collison

Neural data is messy. Thousands of neurons firing in high-dimensional spaces, creating patterns that shift and flow. Traditional analysis tools—correlation matrices, dimensionality reduction, clustering algorithms—capture some of this structure. But they miss something fundamental: shape. Topological Data Analysis (TDA) sees what others can’t. It reveals holes,

What makes a molecule alive? Not what it’s made of—carbon and hydrogen are everywhere. Not even what it does—plenty of complex chemistry happens in dead systems. What makes a molecule biological is how hard it would be to make without selection. This is Lee Cronin’s insight:

Why does asking people “Is Linda a bank teller?” give one answer, but asking “Is Linda a feminist bank teller?” give a different, logically impossible one? Why does the order of questions change people’s responses? Why do human judgments violate classical probability so consistently? For decades, cognitive scientists treated

What if reality isn’t what it seems? Not in a mystical sense, but in a precise, mathematical, evolutionary sense. Donald Hoffman argues that natural selection shaped our perception not to reveal truth, but to hide it. Our senses are like a desktop interface—useful icons that guide action but

Karl Friston’s Free Energy Principle is beautiful theory. But can you actually build something with it? The answer is yes. Active inference—the action-oriented extension of the Free Energy Principle—is moving from blackboards to codebases, from theoretical neuroscience to working AI agents. Researchers are building robots that

For years, AI progress followed a simple law: train bigger models, get better performance. Pour more compute into pretraining, and capabilities scaled predictably. Then OpenAI discovered something that changes everything: you can also scale by thinking harder at inference time. This is test-time compute scaling—the realization that extending